Ken Griffin says AI is ‘almost certainly’ a bubble, but he’s still a believer

Ken Griffin thinks the AI trade has gotten ahead of itself. He also thinks AI is going to change everything.

The Citadel founder and CEO told an audience in New York on October 20, 2026, that investors are “almost certainly” caught in an AI bubble. In the same breath, he made the case that the technology could usher in a golden era for entrepreneurs.

What Griffin actually said

The remarks came during a talk at Citadel’s annual Future of Global Markets Conference. Citadel estimates it manages approximately $68 billion in assets, so his read on market sentiment tends to travel.

His core worry was not that AI is useless. It was that companies are spending heavily and struggling to show what they got for it.

Griffin described corporations taking a hard look at their AI bills, including invoices from providers like Anthropic and OpenAI. Many of those businesses are finding it difficult to point to clear productivity gains from the money they have spent.

Griffin noted that companies are wrestling with both the steep cost of AI and the uncertain timeline for any return on investment.

That tension shows up most clearly in infrastructure spending. US data center spending linked to AI is projected to exceed $500 billion in 2026.

From skeptic to cautious convert

What makes Griffin’s comments interesting is how much his own view has moved. Over the past year, he has shifted from skepticism about AI’s near-term impact toward openly acknowledging its capabilities.

The reason is close to home. Griffin pointed to substantial productivity improvements inside Citadel, driven by agentic AI systems. These are tools designed to carry out multi-step tasks with less hand-holding, rather than simply answering one question at a time.

According to Griffin, Citadel’s internal AI tools have cut the time required for certain high-level finance research from 6-8 weeks down to 2-3 hours.

Griffin suggested AI could open a golden era of entrepreneurship by lowering the barriers for new entrants, which could mean more competition across markets.

A familiar pattern in markets

The market has already shown some nerves. Recent volatility has included a selloff in semiconductor stocks tied to AI infrastructure firms.

What this means for the AI trade

For investors, Griffin’s comments point toward a more selective phase. If companies are scrutinizing AI bills and demanding proof of returns, the easy assumption that every AI dollar spent is a dollar well spent starts to break down.

Infrastructure is the pressure point to watch. With AI-linked data center spending projected to top $500 billion this year, a large share of the market’s AI exposure runs through chips, power, and server capacity. The recent semiconductor selloff offers a preview of how quickly sentiment can turn.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.